CN120832703A

CN120832703A - Process and system for the design of orthotics

Info

Publication number: CN120832703A
Application number: CN202510489399.2A
Authority: CN
Inventors: 张腾; 况熙和
Original assignee: Hong Kong Keluohua Medical Technology Co ltd
Current assignee: Hong Kong Keluohua Medical Technology Co ltd
Priority date: 2024-04-19
Filing date: 2025-04-18
Publication date: 2025-10-24
Also published as: WO2025218774A1

Abstract

The present application relates to a process and system for the design of orthotics. A process operable using a computerized system for providing output data indicative of a geometry of a spinal region of a subject's body for spinal alignment correction, the process comprising the steps of (i) detecting a body landmark of the subject from a three-dimensional point cloud model of a body surface of the spinal region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomy of the subject, (ii) determining a spinal correction of the subject's spine, wherein the spinal correction provides the spinal alignment correction of the subject, and (iii) generating a corrected three-dimensional point cloud model of the spinal region of the subject.

Description

Process and system for the design of orthotics

Technical Field

The present invention relates to the field of orthosis apparatus, and more particularly to a process and system for the design of a spinal orthosis apparatus.

Background

Within the field of orthopedic interventions, there are several aspects, some of which are:

(i) Internal joint replacement or repair devices, commonly referred to as "orthopedic implants" or "prostheses," such as hip, knee and shoulder prostheses,

(Ii) Orthosis devices, such as spinal brackets, are used to intervene in deformities such as scoliosis, and such devices are typically disposed external to and attached to the body, such as spinal brackets.

Within the field of orthopedic and musculoskeletal correction practice, there are also included orthosis devices, which are surgical devices or appliances that generally exert external forces on a portion of the body to support a joint and also correct deformities of a subject.

An example of such an orthosis apparatus is an external correction device, such as a spinal bracket. Such a scaffold is used to apply a force to the spine of a patient, such as a patient suffering from scoliosis.

In particular, children with AIS (adolescent idiopathic scoliosis), a condition that generally affects children between 10 years and adolescence, is indicated by the presence of abnormal curvature of the spine to the right or left of the "S" or "C" shape. Teenagers with scoliosis are generally healthy and are often treated with a stent externally secured to their spine to gradually force the spine through the stent into a more normal state.

In the prior art of design and manufacture of orthosis apparatus such as spinal brackets, typical steps include:

(i) Manually forming a negative mold casting model from the body of the subject;

(ii) Producing positive castings using milling machines typically having polymeric materials;

(iii) Manually correcting the positive die casting, and

(Iv) The polymeric material is manually molded onto the positive mold casting to form the orthosis.

Object of the invention

It is an object of the present invention to provide a process and system for orthosis apparatus design that overcomes or at least partially ameliorates at least some of the disadvantages associated with the prior art.

Disclosure of Invention

In a first aspect, the present invention provides a process operable using a computerized system for providing output data indicative of a geometry of a spinal column region of a subject's body for spinal alignment correction, the process comprising the steps of (i) detecting a body landmark of the subject from a three-dimensional point cloud model of a body surface of the spinal column region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomy of the subject, (ii) determining a spinal column correction of the subject, wherein the spinal column correction provides spinal alignment correction of the subject, and (iii) generating a corrected three-dimensional point cloud model of the spinal column region of the subject, wherein the corrected three-dimensional point cloud model is generated based on the spinal column correction of the subject and a three-dimensional point cloud model of a surface of the spinal column region of the subject, wherein the corrected three-dimensional model comprises output data indicative of a geometry of the spinal column region of the body of the subject including the body landmark of the subject for spinal column alignment correction of the subject, and wherein the three-dimensional point cloud model of the body surface of the spinal column region of the subject and the three-dimensional point cloud model of the spinal column region of the subject are generated during movement of the corrected three-dimensional point cloud model of the spinal column region of the subject.

A three-dimensional point cloud model of a spine region of a subject may be generated from one or more data input sets, wherein each data input set of the one or more data input sets is indicative of an optical image of the subject, and wherein the optical image is a three-dimensional optical image indicative of a geometry of the spine region of the subject. The optical images may be red, green, blue and depth (RGBD) images.

The three-dimensional point cloud model may be generated using a data set from a corresponding one of the three-dimensional optical images of the spine region of the subject. One three-dimensional optical image may be a Posterior Anterior (PA) three-dimensional optical image of a spinal region of the subject.

Three data sets of three corresponding three-dimensional optical images from a spine region of a subject may be used to generate a three-dimensional point cloud model. The three-dimensional optical image of the spinal region of the subject is preferably a Posterior Anterior (PA), left (Lt), and right (Rt) three-dimensional optical image of the spinal region of the subject.

A spinal correction of the subject may be determined from the three-dimensional point cloud model of the spinal region of the subject.

The spinal correction of the subject may be determined from anatomical landmarks of the subject's spine from one or more medical images of the subject's spinal region. The one or more medical images may be one or more X-ray images of a spinal region of the subject.

The one or more medical images may be anterior-posterior (AP) X-ray images of a spinal region of the subject to provide two-dimensional (2D) spinal alignment correction of the subject's spine. The one or more medical images are anterior-posterior (AP) X-ray images and Lateral (LAT) X-ray images of a spinal region of the subject to provide three-dimensional (3D) spinal alignment correction of the subject's spine.

The physical markers of the subject detected from the three-dimensional point cloud model of the surface of the subject's spinal region may be detected by a pre-trained Artificial Intelligence (AI) component. The body landmark positions detected by the pre-trained Artificial Intelligence (AI) component can be further reviewed and fine-tuned as needed by one or more human operators.

A spinal correction for the subject's spinal alignment correction is determined by assigning one or more of torsion, back balance, and spinal curve correction by the one or more human operators.

The corrected three-dimensional point cloud model of the surface of the spinal region of the subject may be provided from the three-dimensional point cloud model and body landmark locations of the surface of the spinal region of the subject by a pre-trained Artificial Intelligence (AI) unit, and may optionally further include fine tuning by one or more human operators by assigning one or more of torsion, back balance, and spinal curve correction.

The output data indicative of the geometry of the spinal region of the surface of the body of the subject for spinal alignment correction is indicative of the geometry of an orthosis for providing the spinal alignment correction to the subject.

In a second aspect, the present invention provides a computerized system for providing output data indicative of a geometry of a spinal region of a subject's body for spinal alignment correction. The system comprises a geometry optimization component for detecting a body landmark of a subject from a three-dimensional point cloud model of a body surface of a spinal region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomy landmark, wherein the geometry optimization is for generating a corrected three-dimensional point cloud model of the spinal region of the subject, wherein the corrected three-dimensional point cloud model is generated based on a spinal correction of the subject and on a three-dimensional point cloud model of a surface of the spinal region of the subject, wherein the spinal correction provides spinal alignment correction of the subject, wherein the corrected three-dimensional model comprises output data indicative of a geometry of the spinal region of the surface of the body of the subject including the body landmark for spinal alignment correction of the subject, and wherein the body landmark of the three-dimensional point cloud model of the body surface of the spinal region of the subject and the anatomical landmark of the spinal column of the subject move during generation of the corrected three-dimensional point cloud model of the spinal region of the subject.

The system may comprise a point cloud generating component for generating the three-dimensional point cloud model of the spine region of the subject from one or more data input sets, wherein each data input set of one or more data input sets is indicative of an optical image of the subject, and wherein the optical image is a three-dimensional optical image indicative of a geometric configuration of the spine region of the subject. The optical images may be red, green, blue and depth (RGBD) images.

A three-dimensional point cloud model may be generated using a data set from a corresponding one of the three-dimensional optical images of the spine region of the subject. One three-dimensional optical image may be a Posterior Anterior (PA) three-dimensional optical image of a spinal region of the subject.

Three data sets of three corresponding three-dimensional optical images from a spine region of a subject may be used to generate a three-dimensional point cloud model. The three-dimensional optical image of the spinal region of the subject may be Posterior Anterior (PA), left (Lt), and right (Rt) three-dimensional optical images of the spinal region of the subject.

A spinal correction of the subject is determined from the three-dimensional point cloud model of the spinal region of the subject.

The spinal correction of the subject may be determined from one or more medical images of a spinal region of the subject. The one or more medical images may be one or more X-ray images of a spinal region of the subject.

The one or more medical images may be anterior-posterior (AP) X-ray images of a spinal region of the subject to provide two-dimensional (2D) spinal alignment correction of the subject's spine.

The one or more medical images may be anterior-posterior (AP) X-ray images and Lateral (LAT) X-ray images of a spinal region of the subject to provide three-dimensional (3D) spinal alignment correction of the subject's spine.

The physical markers of the subject detected from the three-dimensional point cloud model of the surface of the subject's spinal region may be detected by a pre-trained Artificial Intelligence (AI) component.

The system may also include a user interface such that corrected anatomical landmark positions detected by the pre-trained Artificial Intelligence (AI) component are further reviewed by one or more human operators for fine tuning when needed. The spinal correction for the subject's spinal alignment correction may be determined by assigning one or more of torsion, back balance, and spinal curve correction by the one or more human operators.

The system may further comprise a pre-trained Artificial Intelligence (AI) unit, wherein the corrected three-dimensional point cloud model of the surface of the spinal region of the subject is provided by the pre-trained Artificial Intelligence (AI) unit from the three-dimensional point cloud model of the surface of the spinal region of the subject and the body landmark locations.

The system may also include another user interface for fine tuning by one or more human operators by assigning one or more of torsion, back balancing, and spine curve correction.

The system may further comprise an output interface for outputting said data indicative of the geometry of a spinal region of a surface of a body of a subject for spinal alignment correction, the data being indicative of the geometry of an orthosis for providing said spinal alignment correction to the subject.

In a third aspect, the invention also provides a process operable with a computerized system that determines mechanical properties of an orthosis for correction of spinal alignment of a subject. The process comprises the steps of (i) receiving a three-dimensional model of a body surface of a spinal region of a subject and receiving a corrected three-dimensional model of a body surface of a spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and comprises output data indicative of a geometry of the body surface of the spinal region of the subject for spinal alignment correction, (ii) generating a three-dimensional model of an orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject and wherein mechanical properties of the orthosis comprise relative densities, (iii) determining a displacement of points of the corrected three-dimensional model from the three-dimensional model of the body surface of the spinal region of the subject, (iv) determining a strain energy of the orthosis from the displacement at step (iii) and changing a relative density profile of the orthosis until a predetermined threshold of the strain energy is met and until the predetermined threshold of the relative density is met, (v) generating a topology of the orthosis based on the relative density profile of the corrected three-dimensional model of the body surface of the spinal region of the subject and the corrected three-dimensional model until the predetermined threshold of the relative density is met.

The relative density profile of the orthosis is preferably the porosity profile of the orthosis. The porosity distribution may be a non-uniform porosity distribution based on the topological optimization results of the subject.

The three-dimensional model of the body surface of the spinal region of the subject may be a three-dimensional point cloud model, and wherein the corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional point cloud model.

Alternatively, the three-dimensional model of the body surface of the spinal region of the subject may be a three-dimensional mesh model, and wherein the corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional mesh model.

Alternatively, the three-dimensional model of the body surface of the spinal region of the subject may be a three-dimensional volumetric model, and wherein the corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional volumetric model.

The mechanical properties of the orthosis can be determined by the process of the present aspect, and wherein the geometric configuration of the orthosis is based on the corrected three-dimensional model of the body surface of the spinal region of the subject.

At least a portion of the orthosis can be formed by additive manufacturing techniques. At least a portion of the orthosis can be monolithic.

At least a portion of the orthosis can be formed from a polymeric material. At least a portion of the orthosis can be formed of Polyurethane (PE).

In a fourth aspect, the present invention provides a computerized system for determining mechanical properties of an orthosis for correction of spinal alignment of a subject, the system comprising an input interface for receiving a three-dimensional model of a body surface of a spinal region of the subject and for receiving a corrected three-dimensional model of a body surface of a spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and comprises output data indicative of a geometry of the body surface of the spinal region of the subject for spinal alignment correction, and a processor unit for generating a three-dimensional model of the orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein the mechanical properties of the orthosis comprise relative densities, for determining displacements of points of the corrected three-dimensional model from the three-dimensional model, and determining from the displacements of the orthosis energy for changing a relative density distribution of the orthosis until a predetermined threshold of strain energy is met and until the predetermined threshold of energy is met and the predetermined threshold of relative density is met and the relative density is optimized upon the predetermined threshold of the relative density being generated based on the predetermined threshold of the output of the relative density of the orthosis.

Upon satisfaction of a predetermined threshold of relative density, an optimized model of the orthosis may be output, and wherein a geometry of the orthosis is based on the corrected three-dimensional model of the body surface spinal region of the subject.

In a fifth aspect, the invention provides a computerized system for determining mechanical properties of an orthosis for correction of spinal alignment of a subject, the system comprising an input interface for receiving a three-dimensional model of a body surface of a spinal region of the subject and for receiving a corrected three-dimensional model of a body surface of a spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and comprises output data indicative of a geometry of the body surface of the spinal region of the subject for spinal alignment correction, a displacement calculation unit for calculating a displacement of each point in the three-dimensional model of the spinal region of the subject and the corrected three-dimensional model of the spinal region of the subject, a meshing unit for generating a three-dimensional model of the orthosis from the corrected three-dimensional model of the spinal region of the subject, a numerical analysis unit for mechanical simulation of the orthosis, wherein the numerical analysis unit performs the simulation unit based on the three-dimensional model of the orthosis, the displacement of each point in the three-dimensional model and a topology characteristic of the orthosis comprising a relative density, the optimization unit for optimizing the mechanical model based on the strain density and the mechanical model and the predetermined threshold is met when the mechanical density is met.

Alternatively, the three-dimensional model of the body surface of the spinal region of the subject is a three-dimensional volumetric model, and wherein the corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional volumetric model.

The system may further comprise a fine tuning unit for mesh smoothing and gap repair of the mesh.

In a sixth aspect, the present invention provides a corrected orthosis for spinal alignment of a subject, wherein the orthosis has a relative density determined by a process comprising (i) receiving a three-dimensional model of a spinal region of the subject and receiving a corrected three-dimensional model of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and comprises output data indicative of the geometry of the spinal region of the subject for spinal alignment correction, (ii) generating a three-dimensional model of the orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the spinal region of the subject, and wherein mechanical properties of the orthosis comprise relative densities, (iii) determining a strain energy of the orthosis from the three-dimensional model of the displacement at step (iii) and changing the relative density profile of the orthosis until a predetermined threshold for strain energy is met and until a predetermined threshold for relative density is met, wherein the topology of the orthosis is based on the relative density profile of the three-dimensional model of the spinal region of the subject and wherein the geometry of the corrected three-dimensional model of the spinal region of the subject is based on the physical profile of the three-dimensional model.

In a seventh aspect, the present invention provides a process operable using a computerized system for providing output data indicative of a geometry of a spinal column region of a subject's body for spinal column alignment correction and determining mechanical properties of an orthosis for correcting spinal column alignment of the subject, the process comprising the steps of (i) detecting a body landmark of the subject from a three-dimensional point cloud model of a body surface of the spinal column region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomy of the subject, (ii) determining a spinal column correction of the subject, wherein the spinal column correction provides spinal column alignment correction of the subject, (iii) generating a corrected three-dimensional point cloud model of the spinal column region of the subject, wherein the corrected three-dimensional point cloud model is generated based on the spinal column correction of the subject and the three-dimensional point cloud model of the surface of the spinal column region of the subject, wherein the corrected three-dimensional model comprises output data indicative of the geometry of the spinal column region of the subject including the body surface of the subject for spinal column alignment correction of the subject, and wherein the three-dimensional point cloud model of the spinal column region of the subject is generated from the three-dimensional point cloud of the body of the subject during the three-dimensional model of the spinal column region of the subject, and includes output data indicative of a geometry of a body surface of a spinal region of a subject's body for spinal alignment correction, (v) generating a three-dimensional model of an orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein mechanical properties of the orthosis include relative density, (vi) determining a displacement of points of the corrected three-dimensional model from the three-dimensional model of the body surface of the spinal region of the subject, (vii) determining strain energy of the orthosis from the displacement at step (vi) and changing a relative density profile of the orthosis until a predetermined threshold of strain energy is met and until a predetermined threshold of relative density is met, and (vii) generating a topology of the orthosis based on the relative density profile, and upon the predetermined threshold of relative density being met.

In an eighth aspect, the present invention provides a corrected orthosis for spinal alignment of a subject, wherein the geometry of the spinal region of the body of the subject for spinal alignment correction and the mechanical properties of the corrected orthosis for spinal alignment of the subject are determined by the process according to the seventh aspect.

Drawings

In order that a more particular description of the invention briefly described above will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

The drawings presented herein may not be to scale and any reference to dimensions in the drawings or the following description is specific to the disclosed embodiments.

FIG. 1a shows a schematic diagram of an exemplary embodiment of an overall system according to the present invention;

FIG. 1b shows a schematic diagram of another exemplary embodiment of an overall system according to the present invention;

FIG. 1c shows a schematic diagram of yet another exemplary embodiment of an overall system according to the present invention;

Fig. 2a (i) shows a general flow chart of a process of generating a three-dimensional optimization of an orthosis according to the invention;

fig. 2a (ii) shows a general system for generating a three-dimensional optimization of an orthosis according to the invention;

Fig. 2b (i) shows a schematic diagram of an embodiment of an initial 3D model generating component according to the invention.

FIG. 2b (ii) shows a schematic view of an embodiment of a geometry optimizing component according to the present invention;

FIG. 2c (i) shows a top view of a torsion correction according to an embodiment of the present invention;

FIG. 2c (ii) shows a side view of a torsion correction according to an embodiment of the present invention;

FIG. 2c (iii) shows a back-balanced rear view according to an embodiment of the invention;

FIG. 2c (iv) shows a back side balanced rear view according to an embodiment of the invention;

FIG. 2c (v) shows a back-balanced rear view according to an embodiment of the invention;

FIG. 2d (i) shows a rear view of a spinal curve correction according to an embodiment of the invention;

FIG. 2d (ii) shows a side view of a spinal curve correction according to an embodiment of the invention;

FIG. 2d (iii) shows a top view of spinal curve correction according to an embodiment of the invention;

FIG. 3a (i) shows a general flow chart of a process for determining mechanical properties of an orthotic for correction of spinal alignment of a subject, according to the present invention;

FIG. 3a (ii) shows a general system for determining mechanical properties of an orthosis for correction of spinal alignment of a subject in accordance with the present invention;

FIG. 3b (i) shows a flow chart of an embodiment of a process of biomechanically optimizing member according to the present invention;

FIG. 3b (ii) shows a schematic diagram of the embodiment of FIG. 3b (i);

Figure 3b (iii) shows a Representative Volume Element (RVE) in a homogenization simulation;

FIG. 4 is a flowchart representation of an embodiment of the steps of the present invention for optimizing a body scaffold;

FIG. 5 is a flowchart of steps for optimizing a body scaffold of an embodiment of the present invention;

FIG. 6 shows an example of a point cloud processing back surface reconstruction for the embodiment of FIG. 5;

FIG. 7 shows a first step in the process of reconstructing a point cloud from the back surface to an automated grid for the embodiment of FIG. 5;

FIG. 8 shows a representation of a second step of the result of the reconstruction of the back surface of the embodiment of FIG. 5 from a point cloud to an automatic grid;

FIG. 9 shows a third step of the process of re-fitting the grid of sub-regions to the entire surface of the back surface reconstruction of the embodiment of FIG. 5;

FIG. 10 shows a fourth step of the process of fitting control points to the surface of the back surface reconstruction of the embodiment of FIG. 5;

FIGS. 11a and 11b illustrate a process of calibrating all point clouds in an embodiment of the present invention;

FIG. 11c shows a perspective view with respect to the datum plane of FIGS. 11a and 11 b;

FIG. 12 is a graphical representation of the results after 20000 iterations of the final fit plane of the embodiment of FIGS. 11a-11 c;

FIG. 13a shows a perspective view of the results obtained for 2684 human back surface reconstructions and spinal curves in an example of the invention;

FIG. 13b shows a side view of the results obtained for 2684 human back surface reconstructions and spinal curves of FIG. 13 a;

FIG. 13c shows a top view of the results obtained for 2684 human back surface reconstructions and spinal curves of FIGS. 13a and 13 b;

fig. 14 shows an example of correcting the point cloud of the present example;

FIG. 15a is a graphical representation of a spinal curve (Naen) fitted to a spinal curve (Sentin) obtained from the extreme points of the point cloud profile of the present example;

FIG. 15b is a graphical representation of a target spinal curve of the present example;

FIG. 16 is a graphical representation of a modified target point cloud of the present example;

FIG. 17a shows a grid created by ABAQUS for analyzing the present example for static analysis and optimization;

FIG. 17b shows a grid created by the grid created HYPERMESH for static analysis and optimization;

FIG. 18a shows a grid manually adjusted with control zones and points for static analysis and optimization;

FIG. 18b shows a grid with quadrilateral elements manually adjusted at a grid size of 10 for static analysis and optimization;

FIG. 19a shows the calculation of nodal reaction force from a modified spinal curve;

FIG. 19b shows an optimization model with node loading;

FIG. 20a shows the boundary conditions of the top, left and right, and bottom, respectively;

FIG. 20b shows the result of node reaction force extracted as node load;

FIG. 21a shows the results of deformation during static analysis;

FIG. 21b shows the results of stress during static analysis;

FIG. 22a shows the result of deformation during optimization;

FIG. 22b shows the results of stress during optimization;

FIGS. 23a-23d show the results of penalty methods with different iterations for solid isotropic materials according to the present invention;

FIG. 24a shows boundary conditions with a fixed top boundary and static analysis;

FIG. 24b shows the boundary condition with a fixed reference rotation point with reference to FIG. 24 a;

FIG. 25a shows in three dimensions the results of the stress during the static analysis of FIGS. 24a and 24 b;

FIG. 25b shows in three dimensions the results of the deformation during the static analysis of FIGS. 24a and 24 b;

FIG. 26a shows the total strain energy at different minimum densities;

FIG. 26b is a tabular representation of FIG. 26 a;

FIG. 27a shows the volume ratio at different minimum densities;

FIG. 27b is a tabular representation of FIG. 27 a;

28a-28d illustrate material density profiles of different minimum densities;

29a-29d illustrate stress distributions for different minimum densities;

FIGS. 30a-30d illustrate deformation profiles of different minimum densities;

FIG. 31 shows a flow chart of the optimization of the python script in an example of an embodiment of the present invention;

FIG. 32a shows the boundary conditions of the fitting of the elemental material properties (E and nu) in this example;

FIG. 32b shows the undeformed and deformed results of the fitting of the elemental material properties in this example and with respect to FIG. 32 a;

FIG. 32c shows the case generation in the present example and with respect to the fitting of elemental material properties of FIGS. 32a and 32 b;

FIG. 33a shows the result of curve fitting in this example;

fig. 33b shows the result of curve fitting updated with the optimization method. In the present embodiment and referring to fig. 33a;

FIG. 34a shows preliminary results of the relative density of cantilever beams in the 2D case;

FIG. 34b shows preliminary results of stress of cantilever beams;

fig. 34c shows preliminary results of deformation of the cantilever beam;

FIG. 35a shows preliminary results of relative density for a cylindrical element embodiment according to the present invention;

FIG. 35b shows preliminary results of stress according to the present invention and with reference to the example cylindrical element of FIG. 35 a;

Fig. 35c shows preliminary results of a modification of the example cylindrical element according to the invention and with reference to fig. 35a and 35 b;

FIG. 36a shows a flow chart of a python script for model automatic reconstruction in an embodiment of the present invention;

fig. 36b shows an example of preprocessing an ODB file.

Fig. 36c shows an example of a stl file.

FIG. 37a shows a model of a cylindrical rigid body and BC (boundary conditions);

FIG. 37b shows a mesh of the cylindrical rigid body of FIG. 37 a;

FIG. 38a shows the displacement distribution of the cylindrical rigid body of FIGS. 37a and 37 b;

FIG. 38b shows the stress distribution of the cylindrical rigid body of FIGS. 37a, 37b and 38 a-38 c, and

Fig. 38c shows the relative density distribution of the cylindrical rigid bodies of fig. 37a, 37b and 38a and 38 b.

Detailed Description

The present inventors have recognized the shortcomings of the prior art orthoses and, in recognizing the problems of the prior art, have provided a process and system for designing a spinal orthosis apparatus that overcomes the problems of the prior art.

1. Background of the invention

The present inventors have noted limitations of orthoses, such as prior art spinal brackets, and have sought to provide a process and system for designing orthoses that addresses the shortcomings exhibited by prior art orthoses.

It must be understood that while the exemplary embodiments are depicted and described with respect to an orthosis, and in particular a spinal brace, the present invention is not limited to such embodiments of such devices and is applicable to other and alternative orthoses.

2. Correction device

Orthosis apparatus can be used to align and manipulate the body of a subject, however, for purposes of illustration, the most commonly used orthosis in the context of the present invention is a spinal brace.

As known in the art, such stents are attached to a subject for a portion of the day for gradual manipulation and realignment and positioning of the body, particularly the spine of the subject.

2.1 Idiopathic scoliosis

Idiopathic scoliosis is the most common type of spinal deformity that occurs in children at the onset of puberty, and it has a prevalence of up to 5.2%, known as Adolescent Idiopathic Scoliosis (AIS).

The consequences of idiopathic scoliosis include serious developmental problems, serious cardiopulmonary complications, and the like.

2.2. Current method for AIS management

For the treatment of idiopathic scoliosis, decisions are currently made primarily by assessing the severity of spinal deformities assessed by Cobb angle.

Treatment of AIS is typically performed according to severity defined by the Cobb angle, follow-up (mild AIS, cobb angle 10 ° -25 °), non-operative management (moderate AIS, cobb angle 25 ° -45 °), surgery (severe AIS, cobb angle >45 °).

Spinal orthosis management is most common and effective for non-surgical modalities that prevent curve development by providing external corrective forces to the torso of the patient.

2.3. Conventional manual manufacturing methods are

The conventional manual manufacturing method is as follows:

(ii) Using a milling machine to produce positive mold castings from polymeric materials such as polyurethane materials (PE);

(iii) Manually correcting the positive die casting, and

(Iv) Hand molding polymeric material on positive mold casting to form orthosis

2.4. Disadvantages of conventional manual manufacturing methods

Disadvantages of conventional manual manufacturing methods include lack of specific custom precision for complaints from the subject, laborious and time-consuming and costly production, and subject-specific requirements that are not generally specifically adapted.

Furthermore, such orthosis devices (such as spinal brackets) may not necessarily provide the necessary load at the necessary areas on the subject for corrective manipulation of the spine, and excessive load in one location may result in damage to the subject's skin tissue and improper manipulation of the necessary clinical results.

Similarly, an insufficient load on the necessary areas will result in improper, incorrect or sub-standard handling and clinical results.

Disclosure of Invention

In broad aspects, the invention relates to a process and system for designing and optimizing an orthosis-type device.

The process includes utilizing the novel aspects of the present invention which form at least a portion of the device which has been formed and provided so as to have the requisite mechanical structural properties to provide the requisite biomechanical load to the subject.

The main structural element is preferably monolithic and the monolithic structural element is preferably formed by an additive process and is preferably formed of a polymeric material such as Polyurethane (PE).

The monolithic structural element may be considered or alternatively referred to as an "Aperiodic Material Design (AMD)" with a solid isotropic material.

The present invention provides an orthosis-type device having the necessary mechanical properties to meet biomechanical requirements, whereby the orthosis-type device comprises a monolithic structural element for which the design is optimized such that the orthosis-type device has the necessary mechanical properties to provide the necessary biomechanical objectives.

The monolithic structural material elements of such devices are therefore optimized by analytical methods and formed accordingly to meet biomechanical targets and provided in the form of anisotropically graded density lattice materials.

At any portion of the element, the mechanical properties of the monolithic structural element are determined by the relationship between the monolithic structural element's material mechanical properties (modulus of stiffness and failure component) and the anisotropic lattice unit cell architecture (relative density and anisotropy), and are also a function of the monolithic structural element's geometric properties.

It should be noted that the monolithic structural elements used in such orthosis-type devices can be custom designed based on the specific patient requirements that are typically designed for the patient type, as well as having general configurations and designs that are not patient specific but are either size dependent or size independent.

Thus, with respect to embodiments of the inventive aspects of the present invention, it should not be assumed or introduced with respect to application or embodiments limitations, embodiments of the inventive aspects of the present invention are to provide monolithic structural elements having "anisotropic graded density lattice materials" required to meet the requisite biomedical objectives.

In the case of such orthosis apparatus (e.g. spinal brackets), the optimization of the mechanical properties of the monolithic material structural element may be to provide the subject's spine with the necessary load for repair and treatment purposes, such as in the case of patients suffering from scoliosis.

4. Summary of the invention

The present invention provides a process and system for designing and optimizing an orthosis-type device.

Referring to fig. 1a, an exemplary embodiment of an overall system 1000 according to the present invention is provided as shown.

The system 1000 includes:

(i) The initial point cloud component 100,

(Ii) Geometry optimizing component 200, and

(Iii) Biomechanical optimization component 300.

The initial point cloud component 100 receives red, green, blue and depth (RGBD) data indicative of images from different views of a subject, e.g., images that have been acquired by a depth camera, such as by an RGBD camera, e.g., rear front, left and right views (PA, lt, rt) indicative of a spine region of the subject. Such data may be acquired by an image acquisition device such as a depth camera.

The initial point cloud component 100 generates an initial three-dimensional (3D) point cloud of the subject's body from RGBD data.

The geometry optimization component 200 then determines the necessary geometry of the orthosis for the correct alignment of the subject's spine for the therapeutic purpose of the spine alignment correction. This may be done, for example, by Artificial Intelligence (AI) or manually.

The biomechanical optimizing component 300 then optimizes the design of the orthosis so as to achieve the necessary biomechanical parameters to encourage the subject's body to perform the necessary spinal alignment.

The final optimized orthosis model provided by the biomechanical optimizing component 300 can be manufactured, for example, by 3D printing.

As will be appreciated and recognized by those skilled in the art, when corrective measures for interventional corrective alignment of the subject's spine are provided by the spinal orthosis, this is inevitably accomplished in multiple months, typically through multiple alignment steps.

Thus, the geometry optimization component 200 will generally provide a first desired geometry of the subject's spine manipulation and alignment emphasis for a first step in a series of progressive alignment steps.

It will be appreciated that for the first step of spinal alignment, the degree or amount of alignment provided will depend on the subject and clinical parameters associated with that particular subject, and that such decisions are typically made by a spinal expert surgeon or clinician.

When satisfactory progress is made in the alignment of the subject's spine, after the necessary time, the process may be repeated and an updated RGBD image of the subject's spine is acquired and an updated design of the optimized biomechanical orthosis is determined for the next stage of alignment.

The present system may then repeat the process and clinical assessment until a satisfactory clinical result is achieved.

Those skilled in the art will also appreciate that the present invention may suitably select orthotics of predetermined biomechanical characteristics from such a library of orthotics, or may provide customized orthotics manufacture for a particular subject.

Referring now to FIG. 1b, system 1050 is substantially the same as the system of FIG. 1a, however, in addition to receiving RGBD data for a spinal region of a subject, initial point cloud component 100 also receives data indicative of one or more corresponding X-ray images of the subject.

As will be appreciated, one or more corresponding X-ray images should be sufficient to provide the necessary clinical data for determining the geometry of the orthosis to meet the necessary alignment parameters.

Thus, it should be noted that such steps or aspects of utilizing X-ray data of a subject are optional, and in embodiments of the invention, processes and systems may:

(i) Instead of using X-rays, for example with reference to figure 1a above,

(Ii) Only one X-ray image, such as front-to-back (AP) X-rays,

(Iii) For Anteroposterior (AP) and Lateral (LAT) X-rays, more than one X-ray image is used.

In this embodiment, the optimized geometry is determined by both RGBD data and X-ray data of the subject. In other embodiments, the manner in which this may be provided is discussed in further detail below.

Referring now to fig. 1c, another embodiment of the system 1100 of the present invention is shown, wherein the system 1100 comprises the following:

-a first input interface 001 to receive RGBD (red, green, blue and depth) images from an image acquisition device;

-a second input interface I/O002 to receive X-ray images;

Output interface I/O003 for transmitting the optimized orthosis model for subsequent manufacturing, for example by means of a three-dimensional (3D) printer;

-a first user interface UI004 for receiving manual fine tuning and confirmation for landmark and spinal alignment detection from a spinal expert clinician;

-a second user interface UI005 for receiving manual fine tuning and confirmation for geometry optimization from the corrector;

An initial Point Cloud (PC) 3D model generation part 100;

-a geometry optimization component 200;

biomechanically optimized component 300, and

An Artificial Intelligence (AI) component 400.

By way of example, in an embodiment of the present invention, the method of the present invention for use in the above system comprises the steps of:

(i) The first input interface 001 receives red, green, blue and depth (RGBD) images from different views of the subject, e.g., rear, left and right views (PA, lt, rt) from an image acquisition device, indicating a spinal region of the subject;

(ii) The second input interface 002 receives a corresponding X-ray image of the subject. It should be noted that such steps are optional and that in embodiments of the present invention, the processes and systems may (i) not use X-rays, (ii) use only anterior-posterior (AP) X-rays, (3) use both anterior-posterior (AP) and Lateral (LAT) X-rays, and all such embodiments are considered to be within the scope of the present invention;

(iii) An initial Point Cloud (PC) 3D model generation part 100 receives data S1 indicating (RGBD) images from different views of a subject and data S2 indicating X-ray data of the subject, and

(Iv) An initial 3D point cloud of the subject 'S body is generated by an initial Point Cloud (PC) 3D model generation component 100, and data S9 indicative of the initial 3D point cloud of the subject' S body is received by a geometry optimization component 200.

(V) The geometry optimization component 200 then determines the necessary geometry of the orthosis for the correct alignment of the subject's spine for therapeutic purposes of spinal alignment correction.

(Vi) The biomechanical optimizing component 300 then optimizes the design of the orthosis so as to achieve the necessary biomechanical parameters to encourage the subject's body to perform the necessary spinal alignment.

(Vii) The final optimized orthosis model provided by the biomechanical optimization component 300 (e.g., in STL or CAD format), which can be directly 3D printed.

In an embodiment of the invention, the spinal expert clinician provides manual fine tuning and confirmation for the landmark and spinal alignment detection s4 via a first user interface 004 and the corrector provides manual fine tuning and confirmation for the geometry optimization s5 via a second user interface 005.

The Artificial Intelligence (AI) component 400 can optionally be employed for AI landmarks and spinal alignment detection results s6, and for manually fine-tuning and validating landmarks and spinal alignment detection results for further AI model optimization s8.

The Artificial Intelligence (AI) component 400 can optionally be employed for 3D models of the AI geometry optimization results s7 and geometry optimization of the orthosis s 10. Alternatively, this may be done manually or semi-manually, such as a pre-existing database to be accessed.

5. Geometry optimization

Referring to fig. 2a (i), a general flow diagram of a process 200a for generating a three-dimensional optimization of an orthosis according to the present invention is shown.

Process 200a can operate using a computerized system for providing output data indicative of the geometry of a spinal region of a subject's body for spinal alignment correction.

Process 200a includes the steps of:

Step (i) 210a

Detecting a body landmark of the subject from a three-dimensional point cloud model of a body surface of a spinal region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomical landmark of the subject;

Step (ii) 220a

Determining a spinal correction of a spinal column of a subject, wherein the spinal correction provides a spinal alignment correction of the subject, and

Step (iii) 230a

A corrected three-dimensional point cloud model of the spinal region of the subject is generated,

Generating a corrected three-dimensional point cloud model based on the spine correction of the subject and a three-dimensional point cloud model of a surface of a spine region of the subject,

The corrected three-dimensional point cloud model includes output data indicative of a geometry of a spinal region of a surface of a body of a subject including a body landmark of the subject for spinal alignment correction of the subject, an

The body landmark of the three-dimensional point cloud model from the body surface of the spine region of the subject and the anatomical landmark of the spine of the subject move during generation of the corrected three-dimensional point cloud model of the spine region of the subject.

Referring to fig. 2a (ii), a general computerized system 200b for generating three-dimensional optimization of orthotics according to the invention is shown.

Computerized system 200b provides output data indicative of the geometry of the spinal region of the subject's body for spinal alignment correction.

The system 200b includes a geometry optimization component 210b for detecting a body landmark of the subject from a three-dimensional point cloud model of a body surface of a spinal region of the subject, wherein the body landmark is a landmark indicative of a spinal anatomy landmark;

Geometry optimization component 210b is configured to generate a corrected three-dimensional point cloud model of the spine region of the subject,

Generating a corrected three-dimensional point cloud model based on a spine correction of the subject and based on a three-dimensional point cloud model of a surface of a spine region of the subject, wherein the spine correction provides a spine alignment correction of the subject,

The corrected three-dimensional point cloud model includes output data indicative of a geometry of a spinal region of a surface of a body of a subject including the body landmark for spinal alignment correction of the subject, an

A three-dimensional point cloud model of a spine region of a subject may be generated from one or more data input sets, wherein each data input set of the one or more data input sets is indicative of an optical image of the subject, and wherein the optical image is a three-dimensional optical image indicative of a geometric configuration of the spine region of the subject.

Alternatively, the spinal correction of the subject may be determined from anatomical landmarks of the subject's spine from one or more medical images of the subject's spinal region.

The determination of the spinal correction of the subject's spinal column may be accomplished according to a rule-based criteria, which is preferably a clinical assessment criteria. The clinical assessment criteria may be Cobb angle assessment.

The determination of the spinal correction of the subject's spinal column may be performed by a pre-trained Artificial Intelligence (AI) engine. Training of an Artificial Intelligence (AI) engine may be accomplished with evaluation by a clinician (e.g., one or more clinicians), all through a regularly spaced evaluation and learning system, whereby anatomical landmarks are detected by the Artificial Intelligence (AI) engine and recognition of rule-based bone protrusions or body landmarks is applied.

A. Initial three-dimensional (3D) model generation-detail

Referring to fig. 2b (i), an embodiment of a schematic diagram of an embodiment of an initial 3D model generating component 100 according to the present invention is shown.

The initial 3D model generating part 100 may generate an initial 3D point cloud pattern of the human body S9.

(1) RGBD images (R: red, G: green, B: blue, D: depth) from rear front (PA), left (Lt) and right (Rt) views (S1) are acquired from an appropriate image acquisition device.

(2) The RGBD-PC generator 103 can create a 3D point cloud S101 for each acquired RGBD image.

(3) The 3D point clouds from the 3 different views are registered by the Point Cloud (PC) registration unit 104 to generate the 3D point cloud of the human body S102.

(4) The RGBD image from the Posterior Anterior (PA) view is input to a Point Cloud (PC) landmark detector 102, which Point Cloud (PC) landmark detector 102 may detect anatomical landmarks from the RGBD image, typically by using 6 anatomical landmarks, which indicate the C7 vertebra, the lower left scapula angle, the lower right scapula angle, the lower left iliac spine, the posterior right iliac spine and the coccyx tip by determining coordinates of the anatomical landmarks in the RGBD image, and further map the landmarks to a 3D body point cloud based on the coordinates and depth information of each landmark (S6-2).

(5) Anatomical landmarks detected by Artificial Intelligence (AI) S6-2 are further refined and/or validated by human expert S4-2, and the landmarks in the validated 3D point cloud (S8-2) are sent to an alignment registration unit S105 for further processing, and S8 represents the manually refined and validated AI landmarks and spinal alignment detection results.

(6) The X-ray image/images S2 are input to the spinal alignment detector 101, which spinal alignment detector 101 is able to detect the center of each vertebra from the X-ray image by determining the coordinates of the center of the vertebra in the X-ray image. Spinal alignment generally consists of all vertebral centers, these being the midpoints of the vertebral centers from C7 to L5 and the superior endplate of S1.

It must be noted that the invention may or may not utilize the input of X-ray images and thus is the preferred embodiment.

Thus, the system and process pertaining to the present embodiment may be considered to be capable of operating as three (3) different X-ray image inputs S2 as follows:

(i) The spinal alignment detector 101 is not operable or even required to be part of the system;

(ii) Anterior-posterior (AP) only X-ray image input, two-dimensional (2D), spinal alignment detected by spinal alignment detector 101, or

(Iii) Both AP and Lateral (LAT) X-ray image input three-dimensional (3D) spinal alignment is detected by a spinal alignment detector 101.

(7) The spinal alignment, which may be detected by Artificial Intelligence (AI) S6-1, is further refined and/or confirmed by human expert S4-1, and the confirmed spinal alignment S8-1 is sent to alignment registration unit 105 for further processing. The present inventors have developed a deep learning model for spinal alignment detection.

(8) If an X-ray image is input to the system, a spinal alignment (S8-1) is registered to a 3D point cloud of a human body according to a landmark (S8-2) by an alignment registration unit (105) (S102). Both the spinal alignment (S8-1) and the landmarks (S8-2) contain the center locations of C7 and L5 for registration. The final output is a 3D point cloud of the human body with spinal alignment (S9).

If the X-ray image is not input to the system, no spinal alignment is detected and the alignment registration unit (105) is not active, the final output is a 3D point cloud of the human body without spinal alignment (S9).

(9) The Artificial Intelligence (AI) component 400 includes two AI frameworks:

(i) S6-2 for point cloud anatomical landmark detection, and

(Ii) For spinal alignment detection S6-1.

The manually refined and/or validated landmarks (S8-2) and spinal alignment (S8-1) are used to further optimize the AI framework.

B. Geometry optimization using 3D points only (PC-point cloud data)

In an embodiment of the invention, only RGBD data is used, without the obstacle of geometry optimization with X-ray data, see fig. 2b (i):

(i) The RGBD-PC generator 103 will create a 3D point cloud S101 for each view of the RGBD image S1;

(ii) A Point Cloud (PC) registration unit (104) is to register 3D point clouds from different views to generate a 3D point cloud of the body of the subject S102;

(iii) The Point Cloud (PC) landmark detector 102 will detect anatomical landmarks from the RGBD image S1, and the Artificial Intelligence (AI) component 400 can optionally be used for anatomical landmark detection. Alternatively, a human operator (e.g., a healthcare practitioner) may manually determine the location of the marker;

(iv) The registration unit 105 will register the subject S102 and the 3D point cloud of the anatomical landmark S8-2.

In the case of obtaining point cloud data from other methods or processes, steps (i) and (ii) may be omitted and the process begins with step (iii).

C. Geometry optimization using both 3D points (PC-P point cloud data) and X-ray data

For point clouds obtained from, for example, RGBD image acquisition devices:

(ii) The PC registration unit 104 will register the 3D point clouds from the different acquired views to generate a 3D point cloud of the subject S102;

(iii) The PC landmark detector 102 will detect landmarks from the RGBD image S1 and the AI component 400 is used for anatomical landmark detection;

(iv) The spinal alignment detector 101 can detect spinal alignment from X-rays and the AI component 400 is used for alignment detection, and

(V) The registration unit 105 will register the subject S102, the anatomical landmark S8-2 and the 3D point cloud of the spinal alignment S8-1.

Artificial Intelligence (AI) driven landmarks and spinal alignment detection

The current art is generally manual landmarks and spine alignment detection by the spinal surgeon, spinal specialist or clinician.

5.1. Point Cloud (PC) sign detector 102

According to an embodiment of the invention:

a. the present inventors have developed and provided a top-to-bottom AI framework for automatically detecting anatomical landmarks.

B. U-Net++ has been trained to segment body regions from RGBD images.

C. HRNet have been further trained to identify the location of anatomical landmarks from body regions based on RGBD images.

A key advantage provided by this embodiment of the present invention is that it provides faster and more consistent test results.

5.2. Spinal alignment detector 101

In an embodiment of the invention, with respect to the spinal alignment detector 101:

a. the present inventors have developed a top-to-bottom AI framework to automatically detect vertebral centers.

B. the U-Net is trained to segment vertebral areas from X-ray images.

C. HRNet are further trained to identify the location of the 4 corners of each vertebra based on the X-ray images.

D. The vertebral center can be determined by calculating the center positions of the 4 vertebral corners.

6. Geometry optimization

Referring to fig. 2b (ii), a schematic diagram of an embodiment of a geometry optimizing component 200 according to the present invention is shown.

The geometry optimizing component 200 provides:

(A) Manual geometry optimization (i.e., geometry optimization without AI), or

(B) Artificial Intelligence (AI) aided geometry optimization,

As described in the exemplary embodiment of the present invention, a geometry-optimized 3D model S10 of the orthosis is generated for an initial 3D point cloud S9 of the human body.

A. geometry optimization without Artificial Intelligence (AI)

For geometry optimization with only 3D point clouds, the input (S9) will be the 3D point clouds of humans and landmarks.

For geometry optimization with 3D point cloud data and X-ray data, the input S9 will be the 3D point cloud data of the subject, as well as anatomical landmarks and spinal alignment data, and:

(i) The clinician assigns the degree of torsion correction S5-1 to the torsion correction unit 203, which torsion correction unit 203 will torsion correct the subject' S3D point data and the landmarks (and spinal alignment) change accordingly to provide a correction reference;

(ii) The clinician assigns the degree of back balancing S5-2 to back balancing unit 204, which back balancing unit 204 will back balance the subject' S3D point data and the anatomical landmarks (and spinal alignment) are changed accordingly to provide a correction reference, and

(Iii) The clinician assigns the degree of spine curve correction S5-3 to the spine curve correction unit 205, which spine curve correction unit 205 will perform spine curve correction on the subject' S3D point data and the anatomical landmarks (and spine alignment) change accordingly to provide a correction reference.

Thus, the final corrected 3D point cloud S10 is the target of the correction and will be used for further orthosis bracket design.

Rule-based evaluation criteria may be used as clinical evaluation criteria. The clinical assessment criteria may be Cobb angle assessment.

B. Geometry optimization using Artificial Intelligence (AI)

Optimization of geometry for utilization of AI:

(i) Inputting the initial 3D point cloud of the subject, anatomical landmarks (and spine alignment) S9 to the AI geometry optimization unit 202, which will automatically generate an AI geometry optimized 3D model S201;

(ii) Steps (i), (ii) and (iii) of "a". Geometric optimization without Artificial Intelligence (AI) above would also be performed to fine tune the AI-generated results.

AI generation can use 3D generation models to directly generate optimized 3D models.

Alternatively, the AI model may select a particular main stent model from a library based on the patient's 3D point cloud data, anatomical landmarks (and spinal alignment).

Training of the AI model may be accomplished with evaluation by a clinician (e.g., one or more clinicians) through a regularly spaced evaluation and learning system whereby anatomical landmarks are detected by an Artificial Intelligence (AI) engine and recognition of rule-based bone protrusions or body landmarks is applied.

The rule-based evaluation criterion is preferably a clinical evaluation criterion. For example, the clinical assessment criteria may be Cobb angle assessment.

By way of operational example, and with further reference to the embodiment of fig. 2b, geometric configuration optimization with and without AI is demonstrated:

(1) Selection of geometry optimization mode via signal from user S5-0

If the user selects manual mode:

(2) The optimization mode switch 201 will switch up and the initial 3D point cloud S is directly input to the torsion correction unit 203 and the AI geometry optimization unit 202 will not work because it has been bypassed.

(3) The present inventors have developed three (3) unique geometry optimization units:

(a) The torsion correction unit 203,

(B) Back balancing unit 204, and

(C) Spine curve correction unit (205)

Three (3) different geometric transformations are progressively performed on the initial 3D point cloud S9 with the aid of the orthosis for orthosis design.

The orthosis need only assign 1 parameter to each geometry optimization unit S5-1, S5-2, S5-3 to control the degree of transformation of each geometry.

Each parameter S5-1, S5-2, S5-3 has a specific clinical significance.

If the user selects AI auxiliary mode:

The optimization mode switch 201 will switch down and the initial 3D point cloud S9 is input to the AI geometry optimization unit 202, which AI geometry optimization unit 202 will automatically generate an AI geometry optimized 3D model of the orthosis S201.

(4) The AI-optimized 3D model of orthosis S201 is further refined and/or validated by the AN orthosis via 3 unique geometry optimization units torsion correction unit 203, back balancing unit 204 and spine curve correction unit 205 in order to generate a final geometry optimization of orthosis model S10.

The AI component (400) includes an AI framework for automatic geometry optimization of the initial point cloud S9.

The manually refined and/or validated optimization results S10 are used to further optimize the AI framework.

Current orthosis design techniques utilize a three-point compression system to maintain the proper spinal position [1].

Spinal orthoses are classified according to the region of the spine to which they are fixed, such as Cervical Orthoses (CO), cervical Thoracic Orthoses (CTO) and lumbosacral orthoses (LSO) [2, 3].

The design variables are 2D displacements of three predefined points in the system, which are discrete variables and do not guarantee an effective spinal correction and correction efficacy from a 3D perspective.

The current art aims at restoring the normal configuration of the 2D longitudinal spine, which omits the spinal correction from the top view of the human body [4].

6.1 Torsion correction unit, back balancing unit and spine bending correction unit

According to an embodiment of the present invention, there is provided:

a. Torsion correction unit 203

The torsion correction unit 203 provides:

(i) The subject's torso rotation degree was calculated based on PIIS (posterior lower iliac spine) baseline from top view,

(Ii) The degree of rotation of each point in the point cloud is calculated,

(Iii) The midline is found as the axis of rotation (through the centroid),

(Iv) Each point is rotated according to the rotation axis according to the calculated degree S202,

(V) The parameters (ranging from 0% to 100%) S5-1 are manually entered to control the degree of torsion correction (see fig. 2c (i) and 2c (ii)):

A 0% value means that there is no correction,

100% Means that the torso rotation is corrected to 0.

B. back balancing unit 204

The back balancing unit 204 provides:

(i) Freezing the front of the body.

(Ii) The rear of the point cloud is fitted and adjusted according to all points at each height, the goal being to average the bilateral symmetry position of the point cloud S203.

(Iii) Manually inputting parameters (ranging from 0% to 100%) (S5-2) to control the degree of back balance (see FIG. 2c (iii) -2c (iv))

A. a 0% value means that there is no correction,

B. 100% means perfect bilateral symmetry.

C. spine curve correction unit 205

The spinal curve correction unit 205 provides

(I) Calculating the length of the spine curve;

(ii) Fixing the final point of the spinal curve and straightening the curve into a straight line having the same length;

(iii) Moving the point to a corrected position S10 according to the corrected spinal curve, and

(Iv) Manually inputting parameters (ranging from 0% to 200%) S5-3 to control the degree of correction of the spinal curve (see FIGS. 2d (i) -2d (iii))

A. % indicates that there is no correction and,

B. 100% means correcting the spinal curve to a straight line, and

C. 200% means correcting the spinal curve to a symmetrical curve.

Key advantages provided by this embodiment of the invention include:

(i) Spinal deformities are modeled more accurately from a 3D perspective,

(Ii) More accurate and continuous correction of an orthosis model based on 3D correction of a patient's spine, and

(Iii) Spinal correction is performed based on 3D parameters of particular clinical significance.

6.2. Artificial Intelligence (AI) geometry optimization unit

In an embodiment of the invention, regarding the intelligent (AI) geometry optimization unit (202):

a. The deep learning network [5] employing the basic architecture of PointNet is used to extract point cloud features from the entire initial point cloud of the human body (S9).

B. another deep learning network with PointNet architecture is trained to extract multi-scale point cloud features centered on each anatomical landmark and vertebral center.

C. The features from step a and step b are combined by feature concatenation.

D. multilayer perceptrons (MLPs) were developed to generate displacements for each point based on the combined features.

E. The displacement is applied to the initial point cloud to generate a geometry optimized point cloud for orthosis S201.

F. The AI model is trained using a pipeline that generates a countermeasure network (GAN).

Key advantages provided by this embodiment of the invention include:

(i) The novel first AI-driven geometry optimization and design for orthotics,

(Ii) Quick automatic orthosis design, and

(Iii) Consistent optimization results that are not affected by subjective experience or error, etc. of the corrector.

7. Biomechanical optimization

Referring to fig. 3a (i), a general flow chart of a process 300a of determining mechanical properties of an orthosis for correction of spinal alignment of a subject in accordance with the present invention is shown.

Process 300a includes the steps of:

step (i) 310a

Receiving a three-dimensional model of a body surface of a spinal region of a subject and receiving a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of a geometry of the body surface of the spinal region of the subject's body for spinal alignment correction.

Step (ii) 320a

Generating a three-dimensional model of the orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein the mechanical properties of the orthosis include relative density.

Step (iii) 330a

The displacement of the points of the corrected three-dimensional model is determined from the three-dimensional model of the body surface of the spinal region of the subject.

Step (iv) 340a

Determining strain energy of the orthosis from the displacement at step (iii) and changing the relative density profile of the orthosis until a predetermined threshold for strain energy is met and until a predetermined threshold for relative density is met.

Step (v) 350a

And outputting an optimization model of the orthosis when the predetermined threshold of relative density is met.

Fig. 3a (ii) shows a general computerized system 300b for determining mechanical properties of an orthosis for correction of spinal alignment of a subject in accordance with the present invention.

Computerized system 300b includes an input interface 310b for receiving a three-dimensional model of a body surface of a spinal region of a subject and for receiving a corrected three-dimensional model of the body surface of the spinal region of the subject.

The corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of a geometry of a body surface of a spinal region of a subject's body for spinal alignment correction.

The system 300b further comprises a processor unit 320b for generating a three-dimensional model of the orthosis and a numerical mechanical analysis model of the corrected three-dimensional model of the body surface of the spinal region of the subject.

A three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein the mechanical properties of the orthosis include relative density, for:

(i) Determining a displacement of a point of the corrected three-dimensional model from the three-dimensional model, and

(Ii) The method comprises determining strain energy of the orthosis from the displacement, for changing a relative density profile of the orthosis until a predetermined threshold for strain energy is met and until a predetermined threshold for relative density is met, and for generating a topology of the orthosis based on the relative density profile, and upon the predetermined threshold for relative density being met, outputting an optimization model of the orthosis.

The three-dimensional model of the body surface of the spine region of the subject may be a three-dimensional point cloud model, a three-dimensional mesh model, or a three-dimensional volume model, and the corrected three-dimensional model of the body surface of the spine region of the subject is a corrected three-dimensional point cloud model, a three-dimensional mesh model, or a three-dimensional volume model.

Referring to fig. 3b (i), a flow chart of an embodiment of a process of biomechanically optimizing a component is shown, and fig. 3b (ii) shows a schematic diagram of the embodiment of fig. 3b (i) according to the present invention.

The biomechanical optimization component 300 provides biomechanical optimization of the geometry optimization results of the orthosis S10 and generates a final optimized 3D model of the orthosis S3 for further 3D printing and manufacturing of the orthosis.

According to the invention:

(1) The displacement calculation unit 301 calculates the displacement of each point S301 in the point cloud based on the geometry-optimized point cloud S10 and the initial point cloud S9.

(2) The 3D mesh model of geometry optimized orthosis S302 is generated by meshing unit 300 based on the geometry optimized point cloud of orthosis S10.

(3) The displacement S301, the 3D mesh model of the orthosis S302 and the relative density S304-1 are input into the numerical analysis unit 302 for biomechanical simulation of the orthosis S303.

(4) Based on the simulation result S303, the topology optimization unit 304 performs a topology optimization of the orthosis model, which generates an optimized relative density profile of the orthosis model S304-1 and a corresponding porosity design of the orthosis model S304-2.

(5) The optimization terminator 304 will compare the differences between the original and optimized relative density profiles.

If the relative density distribution is not significantly updated and the predetermined threshold is met, the optimization will be terminated and the optimized porosity design of the orthosis model S304-2 will be output.

Otherwise, the optimization will continue and the optimized relative density profile S304-1 will be fed back to the numerical analysis unit 302 for simulating the next iteration of the optimization until a predetermined threshold is met.

The predetermined threshold of the objective function (strain energy) is determined as a termination condition for the optimization problem, which iteratively optimizes the distribution of relative densities until the objective function is less than the predetermined threshold.

(6) The optimized porosity design of the orthosis model S304-2 may be further processed by the fine tuning unit 305 for mesh smoothing and gap repair, which will generate a final optimized 3D model of the orthosis S3 for 3D printing.

Current techniques related to spinal orthosis design currently employ solid orthosis models that are limited by conventional manufacturing methods.

Some designs have employed porous structures in order to improve breathability and comfort. However, in contrast to the present invention, the location of the holes is not mathematically determined or designed to provide the necessary optimized biomechanical results.

Thus, such prior devices do not provide the predetermined necessary biomechanical results and do likely cause localized stress concentrations that are detrimental to the human body during spinal correction. Furthermore, the porosity distribution is not personalized to cater for different biomedical conditions of the patient. Conventional three-point pressure systems do not take into account the topological optimization of the orthosis structure. The structure and shape distribution of the current orthoses is largely dependent on the expertise of the physician, without the need for logic methods, and without the need to consider the necessary biomechanical results.

7.1. Numerical analysis unit

With respect to the numerical analysis unit 302 of an embodiment of the present invention, the embodiment utilizes a multi-scale numerical analysis framework of orthosis-based porous structures to ensure satisfaction of both mechanical constraints of spinal correction and excellent biomedical performance.

Homogenizing:

Homogenization is performed in an elemental basis via discretization of the orthosis model into shell elements.

The porosity profile of the orthosis is controlled via the relative density of each element.

As shown in fig. 3b (iii), a square plate model can be used as a Representative Volume Element (RVE) in the homogenization simulation.

From a macro scale, a homogenized model with different relative densities is obtained via continuous excavation or creation of holes or apertures of different radii at the center of the RVE.

Thus, the compositional relationship between the relative density of the elements and the mechanical properties of the other elements is determined by the homogenization process using the following equation:

Thus, the results of the composition relationship are as follows:

once the relative densities of the elements are determined, other element mechanical properties can be calculated for use as input to further mechanical analysis.

From a macro scale, a physical orthosis model is employed for mechanical simulation using relative densities of elements and other mechanical properties as input information.

Non-uniform porosity distribution:

the present invention provides an analytically determined design method or process for non-uniform porosity distribution based on the topologically optimized results of different subjects in order to meet the necessary biomechanical requirements.

After discretizing the orthosis model into a plurality of rectangular elements, the porosity distribution of the orthosis is controlled via the relative density of each element, which is derived from the results of the topological optimization.

In conjunction with additive manufacturing techniques, an uneven porosity distribution may be achieved in order to provide the necessary orthosis, such as a spinal orthosis.

The advantages of homogenization provided by the present invention include:

Homogenizing:

(i) This ensures a high efficiency of the modeling of orthoses with complex geometric configurations.

(Ii) This greatly improves the computational efficiency in numerical modeling.

Non-uniform porosity distribution:

(i) This ensures a logical and proper material distribution in the orthosis, while avoiding local stress concentrations in the human body, which may lead to discomfort for the subject and in some cases to injury.

(Ii) Custom designs of orthoses with various logic and appropriate porosity distributions may be implemented to accommodate biomedical conditions of a patient.

7.2. Topology optimization unit

The topology optimization unit 304 of the present embodiment may utilize an automated topology optimization framework based on a method of Solid Isotropic Material (SIMP) with penalties to achieve optimal biomedical and structural performance with minimal material.

Objective function:

For the objective function, the total strain energy is Wherein F, U and K represent the global force, displacement, and stiffness matrices, respectively.

In elemental analysis, strain energy can also be expressed as

For a total of N elements。

This optimization problem to obtain the best distribution of relative density is expressed as follows:

Optimizing variables:

the relative density represents the volume ratio of the elements after discretizing the orthosis model, which varies from zero to one.

Relative density ofAnd (3) withAndDifferent, andMust be greater than 0 to avoid singularities in the matrix during computation.

To avoid checkerboard pattern configurations and resulting failure to obtain clear optimization boundaries, convolution methods are required to filter and update the optimized design variables, which can provide self-smoothing results:

the purpose of using a filter function is mainly to avoid too discrete results of the porosity distribution in the actual orthosis.

In the above equation, R represents the radius of the convolution filter, M represents the total number of elements in the domain, and dist (i, j) represents the distance between the two elements.

The filter weights depend mainly on the distance between the discrete elements.

A larger relative distance means that the influence of the interaction is smaller and thus the influence of neighboring elements is reduced.

Conversely, if adjacent elements are very close to each other, their interaction is more influential and the weight needs to be increased.

Constraint conditions:

The volume ratio constraints are as follows:

While The volume of the target optimization orthosis is represented, and V shows the volume of the current iteration step.

The constraint is the volume fraction between the retained material of the target spinal orthosis and the initial modelWhich is defined as 50% in the current optimization process.

An optimization solver:

By solving the control equation ku=f using the Finite Element Method (FEM) to obtain the node displacement and the node force at each node of the structure, the elements are connected by the node, and thus the node force propagates through the node.

Optimization using the Optimality Criteria (OC) method requires calculation of the partial derivatives of the objective function and constraints and by application of lagrangian multipliers [ ]) To solve them.

Wherein U and K are global displacement and stiffness matrices, an、AndIs the element stiffness, displacement and volume.

The objective of the Lagrangian multiplier method is to directly convert the constrained optimization problem to an unconstrained optimization problem by converting the constraints to variables.

The mathematical meaning behind the lagrangian multiplier is the coefficient of each vector in the linear combination of gradients of the constraint equation.Is a scalar quantityIs a vector, andIs a relaxation factor and satisfies the Kuhn-turner condition as follows:

To take into account To simplify the calculation:

Thus, the iterative formula for applying a spinal orthosis update based on the optimality criteria method can be obtained as follows:

Wherein, the Is a damping coefficient that ensures stable convergence. The convergence criterion may be determined based on a difference between maximum components of design variables of two adjacent iteration steps:

Wherein, the Representing an iterative convergence criterion (which is typically set to 0.01). The optimization process is completed when the difference between the maximum values of the design variables in the adjacent analysis steps satisfies the above expression.

The topology optimization provided by the invention has the advantages that:

(i) Topology optimization may be applied to a multi-stage spinal orthosis model, allowing for the generation of a corresponding orthosis for each correction stage;

(ii) It ensures maximum biomedical and structural performance under specific volumetric constraints, resulting in high cost effectiveness of the orthosis product;

(iii) The porosity distribution obtained by topologically optimizing the spinal orthosis is more reasonable, reasonable and appropriate;

(iv) The optimized spinal orthosis avoids localized stress concentrations which can extend the useful life of the orthosis product, and

(V) Automatic optimization greatly improves the design efficiency of the orthosis and helps to eliminate the experience dependence of the physician in the orthosis design.

8. Example of the invention-orthosis

Referring to fig. 4, a flowchart representation 400 of an exemplary embodiment of the steps of the present invention for an orthosis is shown, in this case an optimized body scaffold for a subject's spine.

Step 1-Point cloud data preprocessing (410)

The point data is preprocessed to create a surface model for analysis.

The twist angle and displacement of the spine are obtained from the AI model.

Step 2-stationary State analysis to obtain node RF (420)

Boundary conditions are imported to solve for node RF using infinite elements.

Step 3-optimization based on corrected E and nu (423)

This step includes importing a node RF as a boundary condition for optimizing and updating E and nu in each cycle and obtaining the relative density of each element.

Step 4-automatic reconstruction (440) is generated

This step includes using computer aided design software to automatically reconstruct the stent model and derive the model.

Referring to fig. 5, a flowchart of steps for optimizing a body scaffold of an embodiment of the present invention is shown.

An example of a point cloud processing the back surface reconstruction of the embodiment of fig. 5 is shown in fig. 6. Human back point cloud data is collected and preprocessed. After reconstructing the point cloud of the back surface. As shown in FIG. 6, the entire surface plane is rotating and is then used to refine the spinal curve.

Referring now to fig. 7, there is shown a first step in the process of reconstructing a point cloud of the back surface to an automated grid for the embodiment of fig. 5. The mesh formed is a triangle element mesh of poor quality. Repair of the grid is necessary because many elements overlap and the boundaries are unclear.

Referring now to fig. 8, a representation of a second step of the result of the point cloud to automatic mesh reconstruction of the back surface of the embodiment of fig. 5 is shown.

Referring now to fig. 9, a third step of the process of re-fitting the grid of sub-regions to the entire surface of the back surface reconstruction of the embodiment of fig. 5 is shown. The entire grid is divided into several zones. A grid is created for each partition. The sub-grids in the different regions are then joined together and re-fitted to the entire surface.

Referring now to fig. 10, a fourth step of the process of fitting control points to the surface of the back surface reconstruction of the embodiment of fig. 5 is shown. The control points of the boundary are extracted and divided into four parts. Each portion of the control points is fitted to a plane. Although control points are also extracted and fitted to the surface, the differences between the entire surface and the four planes are compared by boolean.

11 A-22 b, an embodiment of a design optimization procedure for a correction device (i.e., a spinal-support-type correction device) in accordance with the present invention is shown and described.

Referring to fig. 11a and 11b, an example of a process of calibrating all point clouds in an embodiment of the present invention is shown. Fig. 11c shows a perspective view of the reference plane with respect to fig. 11a and 11b, since the reference plane of the initial point cloud is uncertain, an adjustment of the spine needs to be performed on the reference plane, and all point clouds need to be calibrated.

Figure 12 shows a graphical representation of the results after 20000 iterations of the final fit plane of the embodiment of figures 11a-11 c. The reference plane is found by a random sampling consistency method using the point cloud of the calibration plate.

Referring now to fig. 13a-13c, fig. 13a shows a perspective view of the results obtained for 2684 human back surface reconstructions and spinal curves in an example of the invention.

Fig. 13b shows a side view of the results obtained for 2684 human back surface reconstructions and spinal curves of fig. 13 a.

Fig. 13c shows a top view of the results of obtaining 2684 human back surface reconstructions and spinal curves of fig. 13a and 13 b.

Fig. 14 shows an example of correcting the point cloud of the present example. When the point clouds are corrected, some of the point clouds will be lost, so the point clouds will appear to be broken layer by layer. Since the correction of the ridge line is not large, it has no great influence on the point cloud.

Referring now to fig. 15 and 15b, fig. 15a is a graphical representation of a spinal curve (Naen) of the present example fitted to a spinal curve (Sentin) obtained from an extreme point of a point cloud profile.

Fig. 15b shows a graphical representation of the target spinal curve of the present example.

Fig. 16 is a graphical representation of the modified target point cloud of the present example.

An updated spine curve (Naen) is fitted to the spine curve (Sentin) obtained from the extreme points of the point cloud profile. The code is then rewritten for point cloud lateral transformation. Further improvements in code are still needed to mitigate distortion of the low precision point cloud and local points of the high precision point cloud, as shown in fig. 16.

Fig. 17a-18b relate to the preparation of a mesh for an analytical stent according to the present invention.

FIG. 17a shows a grid created by ABAQUS for analyzing the present example for static analysis and optimization, and FIG. 17b shows a grid created by HYPERMESH for static analysis and optimization.

The mesh created by ABAQUS and HYPERMESH has poor quality. The grid elements are unevenly distributed.

FIG. 18a shows a grid manually adjusted with control zones and points for static analysis and optimization.

Fig. 18b shows a grid with manual grid sizing of 10 for static analysis and optimization. The entire model is manually divided into 46 sub-regions to correct the grid node by node and generate the corresponding INP file, resulting in an even distribution of 4 node shell elements.

Fig. 19a-20b relate to calculation of node reaction forces, optimization model of node patterns and boundary conditions of node loads according to the present example.

Fig. 19a shows the calculation of the nodal reaction force from the modified spinal curve. FIG. 19b shows an optimization model with node loading.

Fig. 20a shows the boundary conditions of the top, left and right, and bottom, respectively.

Fig. 20b shows the result of node reaction force extracted as node load.

Fig. 21a-22b relate to a static analysis for determining static deformation and stress and illustrate optimization of deformation and stress according to the present example. Fig. 21a shows the results of deformation during static analysis. Fig. 21b shows the results of stress during static analysis. Fig. 22a shows the result of the deformation during the optimization. Fig. 22b shows the results of the stress during the optimization process.

The optimization process now converges in the second iteration step and the deformation increases from the optimization result, but the deformation profile is the same as the one of the static analysis as shown in fig. 21a and 22 a. The maximum stress is reduced by 64%.

23A-23d, illustrative examples of iterative embodiments of solid isotropic materials according to the present invention are shown.

Fig. 23a shows the first iteration, fig. 23b shows the eighth iteration, fig. 23c shows the 12 th iteration, and fig. 23d shows the 54 th iteration.

For illustrative purposes realized in the practice of the process of the present invention, fig. 24a through 30d provide illustrative examples.

Fig. 24a to 25b show the boundary conditions and the static analysis of the structural cylindrical element.

Fig. 24a shows a static analysis of structural elements and boundary conditions with a fixed top boundary. Referring to fig. 24a, fig. 24b shows a boundary condition with a fixed reference rotation point.

Fig. 25a shows the stress results during static analysis of the elements of fig. 24a and 24b in three dimensions. Fig. 25b shows in three dimensions the results of the deformation of the structural elements of fig. 24a, 24b and 25a during static analysis.

Fig. 26a shows the objective function according to the invention, i.e. the total strain energy at different minimum densities. Fig. 26b is a tabular representation of fig. 26 a.

Fig. 27a shows the volume ratio of different minimum densities based on the volume ratio constraint. Fig. 27b is a tabular representation of fig. 27 a.

Fig. 28a-28d show material density profiles for different minimum densities.

Fig. 28a shows data with a minimum density of 0.1, fig. 28b shows data with a minimum density of 0.3, fig. 28c shows data with a minimum density of 0.5, and fig. 28d shows data with a minimum density of 0.7.

29A-29d illustrate stress distributions for different minimum densities;

Fig. 29a shows data with a minimum density of 0.1 and a maximum stress of 6.283 MPa, fig. 29b shows data with a minimum density of 0.3 and a maximum stress of 124.6 MPa, fig. 29c shows data with a minimum density of 0.5 and a maximum stress of 139.0 MPa, and fig. 29d shows data with a minimum density of 0.7 and a maximum stress of 194.3 MPa.

FIGS. 30a-30d illustrate deformation profiles of different minimum densities;

fig. 30a shows data with a minimum density of 0.1, fig. 30b shows data with a minimum density of 0.3, fig. 30c shows data with a minimum density of 0.5, and fig. 30d shows data with a minimum density of 0.7.

FIG. 31 shows a flow chart of the optimization of the python script in an example of an embodiment of the present invention. In step 3110, models, meshing, and boundary conditions are first created in ABAQUS. Wherein the boundary condition is derived from equal node reaction force loads for the static displacement constraint analysis case. Element and node characteristics 3120 are calculated by the center point of each element and the neighbor weight (Rmin) of each element to avoid the checkerboard.

The data is then imported into ABAQUS for static analysis 3130. The global stiffness matrix need not be modified directly for boundary conditions. If the result converges, then the element density will be appended to the odb file. If the result does not converge, a further step 3160 will be performed. The FOP results are obtained by updating the elemental material properties to calculate elemental sensitivities. These steps are repeated from the static analysis 3130 until the result converges.

Fig. 32a shows the boundary conditions of the fitting of the elemental material properties (E and nu) in this example. Fig. 32b shows the undeformed and deformed results of the fitting in this example and with respect to the elemental material properties of fig. 32 a. Fig. 32c shows the case generation in the present example and with respect to the fitting of the elemental material properties of fig. 32a and 32 b.

Fig. 33a shows the result of curve fitting in this example. Fig. 33b shows the result of curve fitting updated with the optimization method. In this embodiment and referring to fig. 33a. Fig. 34a shows preliminary results of the relative density of the cantilever beams for the 2D case in this example. Fig. 34b shows the preliminary results of the stress of the cantilever beam in this example. Fig. 34c shows the preliminary result of the deformation of the cantilever beam in this example.

Fig. 35a shows preliminary results of the relative density of an example of a cylindrical element of a stent according to the present invention. Fig. 35b shows preliminary results of stresses according to the invention and with reference to the example of cylindrical element of fig. 35 a.

Fig. 35c shows preliminary results of a modification of the example cylindrical element according to the invention and with reference to fig. 35a and 35 b.

FIG. 36a shows a flow chart of a python script for model automatic reconstruction in an embodiment of the present invention. Node coordinates and element relative densities are obtained in the pre-processed ODB file 3610. The center point of each element is calculated by the Heron formula 3620. Holes are created according to a relative density, which is the relationship between E/nu and void ratio. Creating separate solids and extract surfaces. Stl file 3600c is stitched together and exported for 3D printing.

Fig. 36b shows an example of preprocessing an ODB file. Fig. 36c shows an example of a stl file.

Fig. 37a shows a model of a cylindrical rigid body and BC (boundary condition). Fig. 37b shows a mesh of the cylindrical rigid body of fig. 37 a.

Fig. 38a shows the displacement distribution of the cylindrical rigid body of fig. 37a and 37 b. FIG. 38b shows the stress distribution of the cylindrical rigid body of FIGS. 37a, 37b and 38a, and

The introduction and adoption of computer aided design and computer aided manufacturing (CAD/CAM) systems facilitates more accurate digital designs and higher productivity, replacing negative die castings and manual correction procedures.

Additive manufacturing (3D printing) can more accurately manufacture stents to achieve a perfect fit and enable greater customization of the patient.

Aperiodic Material Design (AMD) with Solid Isotropic Material and Penalty (SIMP) can achieve optimal material distribution and achieve stents with higher strength under specific objective functions and constraint conditions.

As provided by the present invention, there are systems and devices that use depth sensing and SIMP to generate AI facilitation with an effective orthosis of AMD for AIS patients. Benefits provided would include effective correction and comfort of wear and use.

9. Reference to the literature

[1] Wepner, justin l. and Alan p.alfano. "principle and composition of spinal orthosis (PRINCIPLES AND Components of Spinal Orthoses)", training patterns and aids. Elsevier,2019.69-89.

[2] Lumsden, r.m.i.i. and Morris, j.m. "in vivo study of axial rotation and immobilization of lumbosacral joints （An in vivo study of axial rotation and immobilization at the lumbosacral joint）",JBJS 50.8（1968）：1591-1602..

[3] Newman, meredith, CATHERINE MINNS Lowe and Karen barker, "spinal orthosis for treatment of vertebral osteoporosis and osteoporotic vertebral fracture: systems review （Spinal orthoses for vertebral osteoporosis and osteoporotic vertebral fracture: a systematic review）", Physics and rehabilitation archives 97.6 (2016): 1013-1025".

[4] Weinstein, stuart L et al, "adolescent idiopathic scoliosis (Adolescent idiopathic scoliosis)", lancet (London, UK) volume 3719623 (2008): 1527-37.doi 10.1016/S0140-6736 (08) 60658-3.

[5] Qi, charles r. Et al, "point net: point set deep learning for three-dimensional classification and segmentation (Pointnet: DEEP LEARNING on point sets for 3d classification and segmentation)" IEEE computer vision and pattern recognition conference corpus.2017.

Claims

1. A process operable using a computerized system for providing output data indicative of the geometry of a spinal region of a subject's body for spinal alignment correction, the process comprising the steps of:

(i) detecting a body landmark of the subject from a three-dimensional point cloud model of the body surface of the spinal column region of the subject, wherein the body landmark is a landmark indicating an anatomical landmark of the spinal column of the subject;

(ii) determining a spinal correction of the spine of the subject, wherein the spinal correction provides a spinal alignment correction of the subject, and

(iii) generating a corrected three-dimensional point cloud model of the spinal region of the subject,

wherein the corrected three-dimensional point cloud model is generated based on the spinal correction of the subject and the three-dimensional point cloud model of the surface of the spinal region of the subject,

wherein the corrected three-dimensional point cloud model includes output data for spinal alignment correction of the subject, the output data indicating the geometric configuration of the spinal region of the surface of the body of the subject, the geometric configuration including the body landmarks of the subject, and

wherein the body landmarks of the three-dimensional point cloud model of the body surface of the spinal region of the subject and the spinal anatomical landmarks of the spine of the subject move during generation of the corrected three-dimensional point cloud model of the spinal region of the subject.

2. The process of claim 1, wherein:

The three-dimensional point cloud model of the spinal region of the subject is generated from one or more data input sets, wherein,

Each of the one or more data input sets is indicative of an optical image of the subject, and

The optical image is a three-dimensional optical image showing the geometric structure of the spinal column region of the subject.

3. The process of claim 2, wherein:

The optical image is a red, green, blue, and depth (RGBD) image.

4. The process of claim 3, wherein:

The three-dimensional point cloud model is generated using a dataset of corresponding three-dimensional optical images of the spinal region of the subject.

5. The process of claim 4, wherein:

The one three-dimensional optical image is a posteroanterior (PA) three-dimensional optical image of the spinal region of the subject.

6. A process according to claim 2 or claim 3, wherein:

The three-dimensional point cloud model is generated using three data sets of three corresponding three-dimensional optical images of the spinal region of the subject.

7. The process of claim 6, wherein:

The three-dimensional optical images of the spinal region of the subject are posteroanterior (PA), left (Lt), and right (Rt) three-dimensional optical images of the spinal region of the subject.

8. A process according to any one of the preceding claims, wherein

The spinal correction of the subject is determined from the three-dimensional point cloud model of the spinal region of the subject.

9. The process according to any one of claims 1 to 7, wherein

The spinal alignment of the subject is determined from the spinal anatomical landmarks of the spine of the subject from one or more medical images of the spinal region of the subject.

10. The process of claim 9, wherein:

The one or more medical images are one or more X-ray images of the spinal region of the subject.

11. The process of claim 10, wherein:

The one or more medical images are anteroposterior (AP) X-ray images of the spinal region of the subject to provide a two-dimensional (2D) spinal alignment correction of the spine of the subject.

12. The process of claim 10, wherein:

The one or more medical images are an anteroposterior (AP) X-ray image and a lateral (LAT) X-ray image of the spinal region of the subject to provide a three-dimensional (3D) spinal alignment correction of the spine of the subject.

13. A process according to any one of the preceding claims, wherein

The body landmarks of the subject detected from the three-dimensional point cloud model of the surface of the spinal region of the subject are detected by a pre-trained artificial intelligence (AI) component.

14. The process of claim 13, wherein:

The body landmark positions detected by the pre-trained artificial intelligence (AI) component are further reviewed by one or more human operators and fine-tuned if necessary.

15. The process of claim 14, wherein:

The spinal correction for spinal alignment correction of the subject is determined by assigning one or more of twist, back balance, and spinal curve correction by the one or more human operators.

16. A process according to any one of claims 1 to 13, wherein

The corrected three-dimensional point cloud model of the surface of the spinal region of the subject is provided by a pre-trained artificial intelligence (AI) unit from the three-dimensional point cloud model of the surface of the spinal region of the subject and the body landmark positions.

17. The process of claim 16, further comprising:

Fine-tuning is performed by one or more human operators by assigning one or more of twisting, back balancing, and spinal curve correction.

18. A process according to any one of the preceding claims, wherein

The output data indicative of the geometry of the spinal region of the surface of the body of the subject for spinal alignment correction indicates the geometry of an orthosis for providing the spinal alignment correction to the subject.

19. A computerized system for providing output data indicative of the geometry of a spinal region of a subject's body for spinal alignment correction, the system comprising:

a geometry optimization component configured to detect a body landmark of the subject from a three-dimensional point cloud model of a body surface of the spinal region of the subject, wherein the body landmark is a landmark indicating an anatomical landmark of the spinal column;

The geometry optimization component is used to generate a corrected three-dimensional point cloud model of the spinal region of the subject,

wherein the corrected three-dimensional point cloud model is generated based on the spinal correction of the subject and based on the three-dimensional point cloud model of the surface of the spinal region of the subject, wherein the spinal correction provides spinal alignment correction of the subject,

wherein the corrected three-dimensional point cloud model includes output data for spinal alignment correction of the subject, the output data indicating the geometric configuration of the spinal region of the surface of the body of the subject including the body landmarks of the subject; and

20. The computerized system of claim 19, further comprising:

A point cloud generating component for generating the three-dimensional point cloud model of the spinal region of the subject from one or more data input sets, wherein

Each of the one or more data input sets is indicative of an optical image of the subject, and wherein,

The optical image is a three-dimensional optical image showing the geometric configuration of the spinal column region of the subject.

21. The computerized system of claim 20, wherein:

The optical image is a red, green, blue, and depth (RGBD) image.

22. A computerized system according to claim 19 or claim 20, wherein:

23. The computerized system of claim 22, wherein:

24. A computerized system according to claim 20 or claim 21, wherein:

25. The computerized system of claim 24, wherein:

26. A computerized system according to any one of claims 19 to 25, wherein:

27. A computerized system according to any one of claims 19 to 25, wherein:

The spinal correction of the subject is determined from one or more medical images of the spinal region of the subject.

28. The computerized system of claim 27, wherein:

29. The computerized system of claim 28, wherein:

30. The computerized system of claim 27, wherein the one or more medical images are an anteroposterior (AP) X-ray image and a lateral (LAT) X-ray image of the spinal region of the subject to provide a three-dimensional (3D) spinal alignment correction of the spine of the subject.

31. A computerized system according to any one of claims 19 to 30, wherein:

32. The computerized system of claim 31 , further comprising:

A user interface enables the corrected anatomical landmark positions detected by the pre-trained artificial intelligence (AI) component to be further reviewed and fine-tuned, if necessary, by one or more human operators.

33. The computerized system of claim 32, wherein:

34. The computerized system of any one of claims 19 to 31 , further comprising:

Pre-trained artificial intelligence (AI) units, where

The pre-trained artificial intelligence (AI) unit provides the corrected three-dimensional point cloud model of the surface of the spinal region of the subject from the three-dimensional point cloud model of the surface of the spinal region of the subject and the body anatomical landmark positions.

35. The computerized system of claim 34, further comprising:

Another user interface for fine-tuning by one or more human operators by assigning one or more of twist, back balance, and spinal curve correction.

36. The computerized system of any one of claims 19 to 35, further comprising:

An output interface for outputting the data indicative of the geometric configuration of the spinal region of the surface of the body of the subject for spinal alignment correction, the geometric configuration indicative of the geometric configuration of an orthosis for providing the spinal alignment correction to the subject.

37. A process operable using a computerized system for determining mechanical properties of an orthosis for correction of spinal alignment of a subject, the process comprising the steps of:

(i) receiving a three-dimensional model of a body surface of a spinal region of the subject and receiving a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of the geometric configuration of the body surface of the spinal region of the body of the subject for spinal alignment correction;

(ii) generating a numerical mechanical analysis model of a three-dimensional model of the orthosis and a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein the mechanical properties of the orthosis include relative density;

(iii) determining, from the three-dimensional model of the body surface of the spinal region of the subject, displacements of points of the corrected three-dimensional model;

(iv) determining a strain energy of the orthosis from the displacement at step (iii) and varying the relative density profile of the orthosis until a predetermined threshold value of strain energy is met and until a predetermined threshold value of relative density is met; and

(v) generating a topology of the orthosis based on the relative density distribution; and outputting an optimized model of the orthosis when the predetermined threshold of relative density is met.

38. The process of claim 37, wherein:

The relative density distribution of the orthosis is the porosity distribution of the orthosis.

39. The process of claim 38, wherein:

The porosity distribution is a non-uniform porosity distribution based on topology optimization results of the subject.

40. A process according to any one of claims 37 to 39, wherein

The three-dimensional model of the body surface of the spinal region of the subject is a three-dimensional point cloud model, and wherein,

The corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional point cloud model.

41. A process according to any one of claims 37 to 39, wherein

The three-dimensional model of the body surface of the spinal region of the subject is a three-dimensional mesh model, and wherein,

The corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional mesh model.

42. A process according to any one of claims 37 to 39, wherein

The three-dimensional model of the body surface of the spinal region of the subject is a three-dimensional volume model, and wherein,

The corrected three-dimensional model of the body surface of the spinal region of the subject is a corrected three-dimensional volume model.

43. An orthosis for correcting spinal alignment of a subject, wherein:

The mechanical properties of the orthosis are determined by a process according to any one of claims 37 to 42, and wherein:

The geometric configuration of the orthosis is based on the corrected three-dimensional model of the body surface of the spinal region of the subject.

44. The orthosis of claim 43, wherein:

At least a portion of the orthosis is formed by additive manufacturing techniques.

45. An orthosis according to claim 43 or claim 44, wherein

At least a portion of the orthosis is unitary.

46. An orthosis according to any one of claims 43 to 45, wherein

At least a portion of the orthosis is formed from a polymer material.

47. The orthosis of claim 46, wherein:

The at least a portion of the orthosis is formed from polyurethane (PE).

48. A computerized system for determining mechanical properties of an orthosis for use in correcting spinal alignment of a subject, the system comprising:

an input interface for receiving a three-dimensional model of a body surface of a spinal region of the subject, and for receiving a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of the geometric configuration of the body surface of the spinal region of the body of the subject for spinal alignment correction; and

a processor unit for generating a numerical mechanical analysis model of the three-dimensional model of the orthosis and the corrected three-dimensional model of the body surface of the spinal region of the subject,

wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and

wherein the mechanical properties of the orthosis include relative density; for determining the displacement of points of the corrected three-dimensional model from the three-dimensional model; and for determining the strain energy of the orthosis from the displacement, for changing the relative density distribution of the orthosis until a predetermined threshold value of the strain energy is met and until a predetermined threshold value of the relative density is met; and for generating a topology of the orthosis based on the relative density distribution and outputting an optimized model of the orthosis when the predetermined threshold value of the relative density is met.

49. The computerized system of claim 48, wherein:

50. The computerized system of claim 49, wherein:

51. A computerized system according to any one of claims 48 to 50, wherein:

52. The computerized system of any one of claims 48 to 50, wherein:

53. The computerized system of any one of claims 48 to 50, wherein:

54. A computerized system according to any one of claims 48 to 53,

When the predetermined threshold value of the relative density is met, the optimized model of the orthosis is output, and wherein,

The geometry of the orthosis is based on the calibrated three-dimensional model of the spinal region of the subject's body surface.

55. A computerized system for determining mechanical properties of an orthosis for correction of spinal alignment of a subject, the system comprising:

an input interface for receiving a three-dimensional model of a body surface of a spinal region of the subject and for receiving a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of the geometric configuration of the body surface of the spinal region of the body of the subject for spinal alignment correction;

a displacement calculation unit configured to calculate a displacement of each point in the three-dimensional model of the spinal column region of the subject and in the corrected three-dimensional model of the spinal column region of the subject;

a meshing unit for generating a three-dimensional model of an orthosis from the corrected three-dimensional model of the spinal region of the subject;

a numerical analysis unit configured to perform mechanical simulation of the orthosis, wherein the numerical analysis unit performs the mechanical simulation based on the three-dimensional model of the orthosis, the displacement of each point in the three-dimensional model, and the mechanical properties of the orthosis including relative density; and

A topology optimization unit for demonstrating a topology optimization of a model of the orthosis based on the results from the numerical analysis unit and for providing an optimized relative density of the orthosis when a predetermined threshold value of strain energy is met and a predetermined threshold value of relative density is met.

56. The computerized system of claim 55, wherein:

57. The computerized system of claim 56, wherein:

58. A computerized system according to any one of claims 55 to 57, wherein:

59. The computerized system of any one of claims 55 to 57, wherein:

60. The computerized system of any one of claims 55 to 57, wherein:

61. The computerized system of any one of claims 55 to 60, further comprising:

Fine-tuning unit for mesh smoothing and gap repair of said mesh.

62. An orthosis for correcting spinal alignment of a subject, wherein the orthosis has a relative density determined by a process comprising:

(i) receiving a three-dimensional model of the spinal region of the subject and receiving a corrected three-dimensional model of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of the geometry of the spinal region of the body of the subject for spinal alignment correction;

(ii) generating a numerical mechanical analysis model of the three-dimensional model of the orthosis and the corrected three-dimensional model of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the spinal region of the subject, and wherein the mechanical properties of the orthosis include relative density;

(iii) determining displacements of points of the corrected three-dimensional model from the three-dimensional model;

(iv) determining a strain energy of the orthosis from the displacement at step (iii) and varying the relative density profile of the orthosis until a predetermined threshold value of strain energy is met and until a predetermined threshold value of relative density is met;

wherein the topology of the orthosis is based on the relative density distribution, and

Therein, the geometric configuration of the orthosis is based on the calibrated three-dimensional model of the body surface of the spinal region of the subject.

63. The orthosis of claim 62, wherein:

64. The orthosis of claim 63, wherein:

65. An orthosis according to any one of claims 62 to 64, wherein

66. An orthosis according to any one of claims 62 to 65, wherein

At least a portion of the orthosis is unitary.

67. An orthosis according to any one of claims 62 to 66, wherein

At least a portion of the orthosis is formed from a polymer material.

68. The orthosis of claim 67, wherein:

At least a portion of the orthosis is formed from polyurethane (PE).

69. An orthosis according to any one of claims 62 to 68, wherein

70. An orthosis according to any one of claims 62 to 68, wherein

71. An orthosis according to any one of claims 62 to 68, wherein

72. A process operable using a computerized system for providing output data indicative of the geometry of a spinal region of a subject's body for spinal alignment correction and for determining mechanical properties of an orthosis for spinal alignment correction of the subject, the process comprising the steps of:

wherein the corrected three-dimensional point cloud model includes output data for spinal alignment correction of the subject, the output data indicating the geometric configuration of the spinal region of the surface of the body of the subject including the body landmarks of the subject, and

wherein the body landmarks of the three-dimensional point cloud model of the body surface of the spinal region of the subject and the spinal anatomical landmarks of the spine of the subject move during generation of the corrected three-dimensional point cloud model of the spinal region of the subject,

(iv) receiving a three-dimensional model of the body surface of the spinal region of the subject and receiving a corrected three-dimensional model of the body surface of the spinal region of the subject, wherein the corrected three-dimensional model is generated from the three-dimensional model and includes output data indicative of the geometric configuration of the body surface of the spinal region of the body of the subject for spinal alignment correction;

(v) generating a numerical mechanical analysis model of a three-dimensional model of the orthosis and a corrected three-dimensional point cloud model of the body surface of the spinal region of the subject, wherein the three-dimensional model of the orthosis is generated from the corrected three-dimensional model of the body surface of the spinal region of the subject, and wherein the mechanical properties of the orthosis include relative density;

(vi) determining displacements of points of the corrected three-dimensional point cloud model from the three-dimensional model of the body surface of the spinal region of the subject;

(vii) determining a strain energy of the orthosis from the displacement at step (vi) and varying the relative density profile of the orthosis until a predetermined threshold value of strain energy is met and until a predetermined threshold value of relative density is met; and

(vii) generating a topology of the orthosis based on the relative density distribution; and outputting an optimized model of the orthosis when the predetermined threshold of relative density is satisfied.

73. An orthosis for correcting spinal alignment of a subject, wherein:

The geometry of the spinal region of a subject's body for spinal alignment correction and the mechanical properties of an orthosis for correcting the subject's spinal alignment are determined by the process of claim 72 .

74. An orthosis according to claim 72 or claim 73, wherein

75. An orthosis according to any one of claims 72 to 74, wherein

At least a portion of the orthosis is unitary.

76. An orthosis according to any one of claims 72 to 75, wherein

At least a portion of the orthosis is formed from a polymer material.

77. The orthosis of claim 76, wherein:

The at least a portion of the orthosis is formed from polyurethane (PE).